OMB Supporting Statement B-rev 8-28-11

SUPPLEMENTAL SUPPORTING STATEMENT B

B. Collection of Information Employing Statistical Methods

Introduction

We intend to use statistical methods in analyzing the data obtained from our two web-based surveys, “E-Verify Survey for Mandatory States” and “E-Verify Survey for Non-Mandatory States.” Analysis will be conducted in each of the three years of the survey and results compared across years. We recognize that there are numerous factors at play between paired non-mandatory and mandatory E-Verify states. The surveys will be used to generate descriptive insights to industry and company experience with E-Verify and provide indications of correlation rather than causation in their results.

Recognizing the numerous different factors at play between paired non-mandatory and mandatory E-verify states, the survey should be used to generate descriptive insights into industry experience.  The comparisons between treatment and control states are unlikely to provide anything more than suggestive evidence about the causal effects of E-Verify, so the study should be extremely cautious in making such claims.

1. Respondent Universe and Selection Methods

The potential respondent universe for the two surveys is all employers in the states of Arizona, Mississippi and South Carolina, where use of E-Verify is mandatory (“mandatory states”) and in the states of Nevada, Alabama and Tennessee, where use of E-Verify may be mandatory for certain employers (e.g. federal contractors) but optional for many employers (“control states”). From that universe, we will select a random sample of approximately 500 companies per state (adjusted to ensure adequate sample size in the various subcategories) that are representative of industries by NCAIS code and number of employees. Our goal from those 500 is to have 150 completed surveys from companies that are representative of the industry and revenue breakdowns appropriate for each state. Within those main categories we also will look at differences in use of E-Verify by: a firm’s annual revenues; urban vs. rural operations; areas with differing levels of unemployment and average household incomes; level of centralization, decentralization or use of outside contractors in the verification process; and sophistication of the human resources function. In addition, there will be significant subsamples from minority (especially Latino) owned businesses and significant subsamples from federal contractors.

The database that will be used to select the random sample comes from each state government’s database of companies in its state (sources are shown below). We will use stratified random sampling.

The total number of firms by state is shown in Exhibit 1a. From that total universe, we will choose a random sample of 500 firms that reflect the breakdown of firms by their industry’s contribution to GDP as shown in Exhibit 2.a.

Exhibit 1.a. Total Universe of Firms by State

State	Number of Firms	Information Source
Arizona	414,109	AZ Dept. of Commerce, 2003
Nevada	322,000	NV Secy. of State, 2008
Mississippi	123,748	US Census County Business Patterns, 2007
Alabama	105,627	US Census County Business Patterns, 2007
South Carolina	99,210	SC Dept. of Commerce 2Q2009
Tennessee	141,744	TN Dept. of Labor & Workforce Dev. 2Q2009

Exhibit 1.b Breakdown of Average Number of Employees

We will select a representative number of businesses using the breakdown by number of employees shown below. This breakdown is commonly used by states analyzing their labor markets.

1-4
5-9
10-19
20-49
50-99
100-249
250-499
500-999
1000-4999
5000-9999

2. Procedures for the collection of information

Statistical methodology

The method for stratification and sample selection will ensure that the random sampling reflects the breakdown of companies in each state in such a way as to mirror the contribution to GDP by industry as well as the percentage of firms by total revenues. The percent contribution by GDP for major industries in each of the six states is show in Exhibits 2.a. through 2.c. below. (Economic profile information was obtained from the U.S. Bureau of Economic Analysis.

Exhibit 2.a. Arizona (Mandatory) compared to Nevada (Control)

ARIZONA			NEVADA
2008 Real GDP* $210,235			2008 Real GDP* $103,192
Key Industry Share of GDP:			Key Industry Share of GDP:
Real Estate Renting & Leasing	16%		Real Estate Renting & Leasing	14%
Retail Trade	8%		Accommodation & Food Services	14%
Health Care & Social Assistance	8%		Construction	8%
Finance & Insurance	7%		Finance & Insurance	8%
Durable Manufacturing	7%		Retail Trade	8%
Professional & Technical Services	6%		Professional & Technical Services	5%
Wholesale Trade	6%		Health Care & Social Assistance	5%
Construction	5%		Wholesale Trade	4%
Admin & Waste Removal Services	4%		Transportation & Warehousing	3%
Accommodation & Food Services	3%		Durable Manufacturing	3%
Government	13%		Government	10%
* Millions of chained $2000 dollars

Exhibit 2.b. South Carolina (Mandatory) compared to Tennessee (Control)

SOUTH CAROLINA			TENNESSEE
2008 Real GDP* $127,065			2008 Real GDP* $210,216
Key Industry Share of GDP:			Key Industry Share of GDP:
Real estate, rental & leasing	11%		Durable Manufacturing (2)	10%
Durable manufacturing	10%		Real estate, rental & leasing (1)	10%
Retail Trade	8%		Health care & social assistance (5)	9%
Nondurable manufacturing	6%		Retail trade (3)	8%
Health care & social assistance	6%		Wholesale trade (6)	7%
Wholesale trade	6%		Nondurable manufacturing (4)	6%
Professional & technical services	5%		Professional & tech services (7)	6%
Construction	5%		Finance & insurance (9)	6%
Finance & insurance	5%		Transp. & warehousing (na)	5%
Administrative & waste services	4%		Admin. & waste services (10)	4%
Government	17%		Government	11%
*Millions of chained $2000 dollars

Exhibit 2.c. Mississippi (Mandatory) compared to Alabama (Control)

MISSISSIPPI			ALABAMA
2008 Real GDP* $71,713			2008 Real GDP* $137,112
Key Industry Share of GDP:			Key Industry Share of GDP:
Durable Manufacturing	9%		Durable Manufacturing	11%
Real Estate Renting & Leasing	8%		Real Estate Renting & Leasing	9%
Retail Trade	8%		Retail Trade	8%
Health Care & Social Assistance	7%		Health Care & Social Assistance	7%
Nondurable Manufacturing	6%		Nondurable Manufacturing	7%
Wholesale Trade	5%		Professional & Technical Services	6%
Construction	5%		Wholesale Trade	6%
Finance & Insurance	4%		Finance & Insurance	5%
Accommodation & Food Services	4%		Construction	4%
Transportation & Warehousing	4%		Transportation & Warehousing	3%
Government	18%		Government	16%
* Millions of chained $2000 dollars

Estimation procedures

We anticipate that our sampling strategy will result in representative samplings of employers by industry type and numbers of employees for each state. To the extent that we are unable to achieve that, we will use weighting to produce statistics that are representatives of the firm breakdown for each state. We also will statistically adjust for non-response as needed.

Degree of accuracy needed

We believe that obtaining 150 completed surveys per state will enable sufficient over-sampling to ensure that results accurately reflect the impact of E-Verify on employers and employees in different industries and different firm sizes. This sample size also allows for attrition over the three-year time frame of the survey (we assume 10 percent attrition over the full three-year time frame), still ensuring that we can make accurate inferences from the data.

Unusual problems

We do not anticipate any unusual problems requiring specialized sampling procedures.

Data collection cycles

We will be collecting data annually for three consecutive years from the same sample.

3. Methods to maximize response rates and deal with issues on non-response

We will devote considerable resources to ensuring adequate response and will continue to survey employers in each state until we achieve the number of responses by employer strata as described earlier. We will follow up each email request with a phone call, and will walk employers through completing the survey to the extent necessary. There will be a help desk open during normal business hours and a web site with FAQs available 24/7.

We anticipate high levels of cooperation among larger employers and employers already using E-Verify, including federal contractors. We will be especially diligent in contacting and helping medium to small employers who we expect will have a less sophisticated human resource function. The survey is designed to be “smart” and only proceed with certain questions based on prior answers. For many companies, therefore, the survey will be relatively simple and quick to complete.

4. Tests of procedures and methods

We will be conducting field tests of the survey as submitted to 10 companies of varying size, type and human resource sophistication to assess the ease of understanding and completing the survey. We will make any technical and wording modifications needed following that test to improve comprehension and ease of use.

5. Contact information for individuals involved

For input on the statistical methodologies used, we contacted:

Craig R. Smith

Assistant Professor

School of Public Administration and Policy

The University of Arizona

(520) 621-4822

Persons involved in the collection, analysis and provision of information to the agency are:

Judith K. Gans

Manager, Immigration Policy Program

Udall Center for Studies in Public Policy

The University of Arizona

(520) 626-9686

Maura Grogan

Technical Expert

Udall Center for Studies in Public Policy

University of Arizona

(520) 975-6274

B-5